Ground beef is made by grinding and mixing raw beef and beef fat. It is the major ingredient in many varieties of high-volume meat products, such as hamburgers and sausages. Consumer marketing of meat products includes the fat content and products are marketed with different lean/fat ratio. High quality protein is also an essential component of meat. Adulteration of meat products is a big problem in the market. This can occur by substituting valuable species of meat with cheaper ones or fresh meat with frozen & subsequently thawed meat. Out-of-spec products are in violation of consumer rights which has legal ramifications. Product recalls for ground beef and other types of food are expensive and in the current social media environment can be fatal to a company’s business and reputation. NIR spectroscopy is a proven method for measuring fat, protein, and moisture in ground beef as well as detecting adulterants.
- Fatty Acids Profile
- Color Measurements
- Adulteration Identification and Quantification
Summary of Published Papers, Articles, and Reference Materials
NIR spectroscopy is an accurate and validated method for measuring fat, protein, and moisture in many food products, including ground beef. There are numerous studies measuring these parameters in ground beef in both laboratory and on-line settings. Fresh, frozen, frozen-thawed, and cooked beef as well as intact beef carcass to a lesser extent have all been studied. There are challenges to building calibration models in intact beef that warrant further study before NIR spectroscopy can be validated as a practical method for measuring these parameters. Multi-point quality measurement has potential as a Process Analytical Technology (PAT) tool for providing real-time feedback for process control. Classification models for turkey meat adulteration in fresh, frozen-thawed, and cooked minced beef samples have validated the use of NIR spectroscopy as a tool for measuring meat adulteration and fraud detection.
Scientific References and Statistics
A Review of the Principles and Applications of Near-Infrared Spectroscopy to Characterize Meat, Fat, and Meat Products – Prieto, Pawluczyk, Dugan, Aalhus, Applied Spectroscopy. 2017, Vol. 71 (7) 1402-1426
Ground Beef – Study 1
Homogenized beef samples scanned in reflectance mode using benchtop laboratory instrument. A limited number of cattle (sixty-three) were used in the study and fed diets containing either sunflower or flax seed for the purposes of modifying fatty acid profiles.
|Fat||R² = 0.86|
|Protein||R² = 0.85|
|Moisture||R² = 0.90|
|Fatty Acids Profile:|
|Saturated Fatty Acids (SFA)||R² = 0.97|
|Monounsaturated Fatty Acids (MUFA)||R² = 0.96|
|Polyunsaturated Fatty Acids (PUFA)||R² = 0.96|
Results were adequate for screening purposes for fat, protein, and moisture; this most likely occurred due to the limited number of cattle used in the study. An excellent correlation for SFA and MUFA were obtained but the correlation for PUFA was poor; this result occurred due to insufficient variability in the data and/or low concentrations. However, if predictions are accurate for SFA and MUFA, then the PUFA concentration can be obtained from the difference between the total Fatty Acids and the sum of SFA and MUFA.
Ground Beef – Study 2
182 homogenized beef samples scanned in reflectance mode and the sample set was designed to incorporate a large range of variability for the parameters of interest.
|Fat||R² = 0.998|
|Protein||R² = 0.99|
|Moisture||R² = 0.99|
The difference in results between these two studies is proof of the importance of creating variability in calibration models. Study two proves the feasibility of accurate fat, protein, and moisture measurements using NIR spectroscopy and a good calibration data set.
On-line Prediction of Chemical Composition of Semi-Frozen Ground Beef by Non-Invasive NIR Spectroscopy – Togersen, Arnesen, Nilsen, Hildrum, Meat Science 63 (2003) 515-523
Semi-frozen Ground Beef (On-Line)
A filter-based non-contact NIR spectrometer was mounted at the outlet of a meat grinder and calibration models were built for fat, protein, and moisture. Fifty-five beef batches of 400 kg to 800 kg in the range of 7.66% to 22.91% fat, 17.04% to 20.76% protein and 59.36% to 71.48% moisture were ground through 4 mm or 13 mm hole plates before scanning
|Fat||R² = 0.97|
|Protein||R² = 0.80|
|Moisture||R² = 0.96|
Statistics are for combined calibrations using the 4 mm and 13 mm samples and there was uneven distribution between the two groups. Good results for fat and moisture (which have a direct correlation) but the protein results were a lot worse. This likely occurred because of a small range in the calibration data and error in the reference method. https://www.sciencedirect.com/science/article/abs/pii/S0309174002001134
On-line Prediction of Beef Quality Traits Using Near Infrared Spectroscopy – Massimo De Marchi, Meat Science 94 (2013) 455-460
Two trials conducted on two hundred thirty young bulls and beef heifers. A fiber optic probe was applied directly to the carcass surface to collect visible-near infrared (VIS-NIR) spectra.
|pH||SECV = 0.04|
|L (Lightness)||SECV = 1.67|
|a (Redness)||SECV = 1.33|
|b (Yellowness)||SECV = 0.96|
|H (Hue Angle)||SECV = 3.28|
|SI (Saturation Index)||SECV = 1.66|
|Cooking Loss %||SECV = 1.79|
|WBSF (Shear Force)||SECV = 6.51|
Measuring pH shows satisfactory results for carcass beef despite a very small range of values in both trials. Most color measurements are feasible as well. Cooking loss is not measured directly but the wavelengths of interest are associated with fat, protein, and moisture so an indirect correlation was obtained. In this study, the authors found that shear force was not measurable using NIR spectroscopy under the parameters used. However, other studies have shown the feasibility of this measurement.
Challenges in Model Development for Meat Composition Using Multipoint NIR Spectroscopy from At-Line to In-Line Monitoring – Dixit, Casado-Gavalda, Cullen, Sullivan, Journal of Food Science, Vol. 82, Nr. 7, 2017
Comparison of At-Line and On-Line Fat and Moisture in Ground Beef
Minced lean beef and beef fat trimmings were scanned using a NIR reflectance spectrometer for the purpose of creating calibration models for fat and moisture. Samples were first scanned under static conditions to simulate an at-line scenario and then scanned under motion conditions to simulate an on-line scenario. Four different probes were used and five different regions were scanned for each sample for twenty total measurements per sample.
|Fat||R² = 0.97, SEP = 6.84%|
|Moisture||R² = 0.98, SEP = 4.72%|
|Fat||R² = 0.97, SEP = 5.95%|
|Moisture||R² = 0.96, SEP = 4.33%|
The on-line Standard Error of Prediction for both parameters was slightly better for the on-line predictions than for the at-line predictions. This is likely because the probes scan a larger sampling area for the on-line spectra collection. The results here prove the feasibility of using on-line measurements at multiple process points for measuring fat and moisture in ground beef.
Identification and Quantification of Turkey Meat Adulteration in Fresh, Frozen-Thawed, and Cooked Minced Beef by FT-NIR Spectroscopy and Chemometrics – Alamprese, Amigo, Casiraghi, Engelsen, Meat Science 121 (2016), 175-181
Turkey Adulteration in Fresh, Frozen-Thawed, and Cooked Minced Beef Samples
Eleven different batches of beef bottom round meat and eleven batches of turkey breast meat were minced separately and used to prepare mixtures with different percentages of turkey meat. All mixtures were scanned as well as the pure beef and turkey. All samples were frozen for six months, thawed, and scanned. After thawing, the samples were cooked in a microwave, cooled, and scanned once again.
|Fresh||R² = 0.925||RMSEC = 8.09|
|Frozen-Thawed||R² = 0.898||RMSEC = 9.39|
|Cooked||R² = 0.916||RMSEC = 8.46|
Both Partial Least Squares (quantification) and PLS-DA (classification) models were used for this study. Results show the potential of using NIR spectroscopy as a reliable tool for the rapid identification and quantification of turkey adulteration in all three types of samples. The classification model can distinguish between adulteration levels less than & greater than 20%. While the quantification models cannot measure the adulteration level if it is less than 20%, this is not of practical importance in a real-time setting.